import type { StructuredProcessOptions, UnstructuredProcessOptions } from './types' export const DEFAULT_GENERATION_PROMPT = '你是一名专业的数据生成助手。请根据输入内容生成准确、完整、可直接用于模型训练的问答数据。仅输出符合目标格式的内容,答案应事实清晰、语言自然,不要添加分析过程、说明或无关内容。' export function createDefaultStructuredOptions(): StructuredProcessOptions { return { preprocessOptions: ['clean_invalid', 'detect_structure', 'deduplicate', 'normalize_format'], semanticEnrichment: false, qaPairsPerRow: 1, datasetSplit: { train: 80, validation: 10, test: 10 }, generationModelId: '', generationPrompt: DEFAULT_GENERATION_PROMPT, temperature: 0.7, maxTokens: 1024, jsonMode: false, qualityFilterEnabled: false, filterLowQuality: true, filterShortContent: true, minOutputLength: 20, } } export function createDefaultUnstructuredOptions(): UnstructuredProcessOptions { return { preprocessOptions: [ 'clean_invalid_content', 'detect_document_structure', 'merge_short_content', 'filter_low_quality', 'deduplicate_content', 'preserve_context', ], chunkMethod: 'structure', chunkSize: 800, chunkOverlap: 100, minChunkSize: 100, customDelimiter: '', preserveTables: true, preserveCodeBlocks: true, preserveLists: true, semanticEnrichment: false, qaPairsPerChunk: 1, datasetSplit: { train: 80, validation: 10, test: 10 }, generationModelId: '', generationPrompt: DEFAULT_GENERATION_PROMPT, temperature: 0.7, maxTokens: 1024, jsonMode: false, qualityFilterEnabled: false, filterLowQuality: true, filterShortContent: true, minOutputLength: 20, } }